Meta-Analyses of Blood Pressure Lowering Trials and the Blood Pressure Lowering Treatment Trialists’ Collaboration


Acknowledgments

The authors would like to thank members of the Blood Pressure Lowering Treatment Trialists’ Collaboration: L. Agodoa (National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland); C. Anderson, J. Chalmers, S. MacMahon, A. Rodgers, and B. Neal (George Institute, Sydney, Australia); F. W. Asselbergs and W. H. van Gilst (University of Groningen, Groningen and Medical Centre Utrecht, Netherlands); C. Baigent and R. Collins (Clinical Trial Service Unit, University of Oxford, Oxford, United Kingdom); E. Berge and M. Mehlum (Oslo University Hospital, Oslo, Norway); H. Black (New York University School of Medicine, New York); F. Bouwers (University Medical Center Groningen, Groningen, Netherlands); B. Brenner and M. Pfeffer (Brigham and Women’s Hospital, Boston, Massachusetts); C. Bulpitt and P. Poole-Wilson (Imperial College, London); R. Byington (Wake Forest University, Winston-Salem, North Carolina); J. Cutler (National Heart, Lung and Blood Institute, Bethesda, Maryland); B. Davis (University of Texas School of Public Health, Houston, Texas); D. de Zeeuw (University Medical Center Groningen, Groningen, Netherlands); J. Dens (University Hospital Gasthuisberg, Leuven, Belgium); R. Estacio (University of Colorado Health Sciences Center, Denver); R. Fagard (University of Leuven, K U Leuven, Belgium); K. Fox (Royal Brompton Hospital and Imperial College, London, United Kingdom); T. Fukui (St. Luke’s International Hospital, Tokyo); L. Hansson and R. Holman (Oxford Centre for Diabetes, Endocrinology and Metabolism, University of Oxford, Oxford, United Kingdom); Y. Imai and T. Ohkubo (Tohoku University Graduate School of Pharmaceutical Sciences and Medicine, Sendai, Japan); M. Ishii (Yokohama Seamen’s Insurance Hospital, Yokohama, Japan); Y. Kanno and H. Suzuki (Musashino Tokusyukai Hospital, Tokyo, Japan); S. Kjeldsen (Ullevaal University Hospital, Oslo, Norway);

J. Kostis (UMDNJ-Robert Wood Johnson Medical School, New Brunswick, New Jersey); K. Kuramoto (Tokyo Metropolitan Geriatric Hospital, Tokyo); J. Lanke (Lund University, Lund, Sweden); E. Lewis (Rush University Medical Center, Chicago); M. Lièvre (Louis Pradel Hospital Université Claude Bernard-Lyon 1, Lyon, France); L.H. Lindholm (Department of Public Health and Clinical Medicine, Umeå University, Umeå, Sweden); L. Lisheng (Fu Wai Hospital and Cardiovascular Institute, Beijing, China); J. Lubsen (SOCAR Research S.A, Nyon, Switzerland); S. Lueders and J. Schrader (St. Josefs Hospital, Cloppenburg, Germany); E. Malacco (Ospedale L. Sacco, University of Milan, Milan, Italy); G. Mancia (University of Milano-Bicocca, Department of Clinical Medicine and Prevention, San Gerardo Hospital, Milan, Italy); M. Matsuzaki (Yamaguchi University Hospital, Yamaguchi, Japan); S. Nissen (Cleveland Clinic, Cleveland); H. Ogawa (The Heart Institute of Japan, Tokyo Women’s Medical University, Tokyo, Japan); T. Ogihara (Osaka University Graduate School of Medicine, Osaka, Japan); T. Ohkubo (Teikyo University, Tokyo Japan); C. Pepine (University of Florida, Gainesville, Florida); B. Pitt (University of Michigan School of Medicine, Ann Arbor, Michigan); M. Rahman (University Hospitals of Cleveland Case Medical Center, Cleveland); H. Rakugi (Osaka University Graduate School of Medicine, Osaka Japan); W. Remme (Sticares Cardiovascular Research Institute, Rhoon, Netherlands); G. Remuzzi and P. Ruggenenti (Mario Negri Institute for Pharmacological Research and Ospedali Riuniti di Bergamo, Bergamo, Italy); T. Saruta (Keio University School of Medicine, Tokyo, Japan); R. Schrier (University of Colorado School of Medicine, Denver); P. Sleight (University of Oxford and John Radcliffe Hospital, Oxford, United Kingdom); J. Staessen (University of Leuven, Leuven, Belgium); K. Teo (McMaster University Medical Centre, Ontario, Canada); L.Thijs (University of Leuven, Leuven Belgium); K. Ueshima (Kyoto University Graduate School of Medicine, Kyoto, Japan); S.Umemoto (Yamaguchi University, Yamaguchi, Japan); P. Verdecchia (Hospital of Assisi, Assisi, Italy); G. Viberti (King’s College London, Guy’s Hospital, London); J. Wang (Ruijin Hospital, Shanghai); P. Whelton (Loyola University Medical Center, Maywood, Illinois); L. Wing (Flinders University, Adelaide, Australia); Y. Yui (Kyoto University Hospital, Kyoto, Japan); S. Yusuf (Hamilton General Hospital, Ontario, Canada); A. Zanchetti (University of Milan and Instituto Auxologico Italiano, Milan, Italy).

BPLTTC Steering Committee: K. Rahimi–Chair (The George Institute for Global Health UK, University of Oxford, Oxford, United Kingdom); J. Chalmers (The George Institute for Global Health, Sydney Australia), B. Davis (University of Texas School of Public Health, Houston, Texas); C. Pepine (University of Florida, Gainesville, Florida); K. Teo (McMaster University, Ontario, Canada).

Over the last few decades, meta-analyses have been central to the advancement of knowledge in a broad range of medical specialties. The “conscientious, explicit and judicious use” of the evidence provided by this technique now underpins much of clinical practice and allows clinicians to make truly informed decisions about how best to deliver care to many different types of patients. In the field of cardiovascular (CV) disease, meta-analyses of the effects of different blood pressure (BP) lowering regimens have allowed for the integrated interpretation of the effects of different therapeutic approaches, and have provided precise estimates of the effects of BP lowering on major CV events, including stroke and coronary heart disease (CHD). As a result, practitioners are now better informed about the implications of their choices of BP-lowering treatment than almost any other mode of therapy to which they have access. For example, meta-analyses have made it possible to determine whether or not important differences exist between drug classes in the protection they afford against different types of serious CV events, and to identify whether the benefits obtained vary according to important characteristics of patients such as risk, age, gender, and the presence or absence of underlying disease. This chapter outlines some key features of meta-analyses and reports the main findings from the most recent, large meta-analyses, including those from the Blood Pressure Lowering Treatment Trialists’ Collaboration (BPLTTC).

Meta-Analyses

The term meta-analysis describes the statistical procedure whereby the results of several different studies addressing the same or related question are combined in an effort to obtain a more precise and more reliable answer to the question under investigation. The technique may be used for quantitatively summarizing data from a range of different study designs (both observational and interventional), usually through identification of relevant studies in a systematic review of the literature. Meta-analyses of randomized controlled trials have been particularly useful because, although the individual estimates provided by small or modest sized trials may be imprecise, the estimates are usually not biased, as long as the individual trials are properly conducted. Thus the combined result of relevant, high-quality randomized controlled trials should give both a more precise and accurate estimate of the real effect of the intervention under investigation, when compared with the findings from individual trials. In addition to providing a more reliable answer to the original research question posed by individual trials, and by providing clarity in fields where there may be individual studies that may appear to be inconsistent, large meta-analyses often have the statistical power to go beyond those original questions by investigating complementary questions relating to treatment effects in important patient subgroups or on less commonly investigated outcomes.

Ultimately, any meta-analysis will have differences in the characteristics of the included trials; for example, trials addressing the effects of different BP-lowering regimens on major CV events have frequently been combined, but included quite varied participants and markedly different durations of follow-up. Likewise, there are many trials investigating the effects of regimens based on one drug class compared with another, but the specific drugs used and the dosing regimens used vary among them. Whether such differences in trial characteristics ultimately strengthen or weaken meta-analysis findings has been the topic of considerable discussion. On balance, it appears that the availability of multiple different studies with different characteristics probably strengthens, rather than weakens, the conclusions. In particular, exploration of the constancy of treatment effects across different participant subgroups and different trial groupings can be done, if a range of similar but not identical trials is included.

The value of meta-analyses depends on the quality and scope of the individual trials included in them. To obtain unbiased estimates of the treatment effect in a meta-analysis of randomized controlled trials, it is essential that the trials included in the meta-analysis are individually and collectively unbiased. It is well established that trials with inconclusive or unfavorable results are not published as frequently as trials with positive findings (i.e., publication bias), and the systematic exclusion of unpublished neutral or negative trials could result in effect estimates from a meta-analysis being biased toward a positive result. Meta-analyses based solely on published data and done without the cooperation of industry or lead investigators in the field are relatively easy to conduct, but may be especially prone to publication bias. By contrast, more resource-intensive meta-analysis projects conducted by large, well-informed collaborative networks, are less subject to publication bias. Example of such collaborative meta-analyses are those conducted by the Blood Pressure Lowering Treatment Trialists’ Collaboration, the Cholesterol Treatment Trialists’ Collaboration, and the Antithrombotic Treatment Trialists’ Collaboration. The prospective and comprehensive nature of such projects limits the potential for bias, because major decisions about analysis and reporting are often specified before the results of any of the contributing trials are known or before pooled analyses are conducted, and major efforts by the broad collaborative group ensure that all relevant trials are identified. With strong collaborative arrangements, there is also considerably enhanced scope for the standardization of outcome definitions and the sharing of individual patient-specific datasets with consequent analytic advantages.

The following sections outline the findings from large meta-analyses of different BP-lowering regimens, which have helped shape our knowledge of their effects on major CV events.

The Blood Pressure Lowering Treatment Trialists’ Collaboration

The BPLTTC is an international collaboration involving the principal investigators of large randomized trials of BP-lowering regimens. The collaboration was established in 1995 with the broad aim of providing the most reliable evidence possible about the effects of commonly used BP-lowering drugs on major CV events using prospective meta-analyses of randomized trials. The meta-analyses are all conducted and reported in accordance with protocols that prespecify research questions, trial eligibility criteria, outcomes, main treatment comparisons, and analysis plans.

Trials are eligible for inclusion in the BPLTTC if they satisfy one of the following criteria: (1) random allocation of patients to regimens based on different BP-lowering agents, (2) random allocation of patients to a BP-lowering agent or placebo, or (3) random allocation of patients to various BP goals. In addition, eligible trials must have a (planned) minimum follow-up of 1000 patient-years per treatment arm. Although trials with factorial assignment to other interventions such as cholesterol-lowering treatment are eligible for inclusion, trials in which additional treatments are jointly assigned with BP-lowering treatment are not eligible, as these other treatments act as potential confounders. For the initial cycle of the collaboration, trials could not have published or presented main trial results before the establishment of the Collaboration in 1995. However, more recently the collaboration has widened its scope with the aim of addressing some of the key remaining questions relating to safety and efficacy of blood pressure lowering.

One key feature of the collaboration is that it gathers individual participant level data from each participating trial wherever possible. The three key advantages of such individual participant data meta-analyses are (1) the benefit of carrying out detailed data checking, (2) the opportunity to better stratify participants into important subgroups using a consistent approach across trials, and (3) the possibility of time-to-event analysis, which increase statistical power, and standard tabular meta-analyses based on aggregate participant data are unable to provide. As per initial agreements, the data requested from investigators included participant characteristics recorded at screening or randomization, selected measurements made during follow-up, and details of the occurrence of all prespecified outcomes during the scheduled follow-up period. In the third cycle of the data collection, which commenced in 2014, all new and existing collaborators were asked to share the full trial dataset, if possible, to facilitate a series of new analyses relating to safety and efficacy of blood pressure lowering.

Since its establishment, the BPLTTC has reported the findings of the overall effects of different BP-lowering regimens in a broad range of patients at risk of CV disease, as well as the effects in specific patient subgroups classified according to patient age, gender, baseline BP, baseline CV risk, and presence or absence of diabetes mellitus (DM). In parallel, there have been other large-scale meta-analyses of BP lowering which have complemented the evidence-base that has been generated by the collaboration.

Overall Effects of Blood Pressure Lowering Among High-Risk Patients with Elevated Blood Pressure

The second cycle of BPLTTC reported updated overall effects of different BP-lowering regimens on major CV events based on data from 29 trials and nearly 160,000 patients. In the majority of trials, patients were selected on the basis of high BP and an additional CV risk factor such as DM, renal disease, or increased age. The overall mean age of participants was 65 years, and just over half (52%) were men. The mean duration of follow-up for contributing trials ranged from 2.0 to 8.4 years, resulting in over 700,000 patient-years of follow-up.

This analysis showed that, compared with placebo, significant reductions in the risk of stroke (28%-38%) and CHD (22%) could be achieved with regimens based on angiotensin-converting enzyme (ACE) inhibitors or calcium channel blockers (CCBs) ( Fig. 47.1 ). In trials that randomized patients to receive either more intensive (lower BP targets) or less intensive BP-lowering regimens, there was also a significant reduction in stroke and a nonsignificant trend toward benefit for CHD with more intensive BP reduction. Heart failure (HF) events were defined as those resulting in death or admission to hospital, and the overviews demonstrated a protective effect against these events from regimens based on ACE inhibitors compared with placebo (18%), a nonsignificant trend toward harm for CCB-based regimens, and a nonsignificant trend toward benefit for regimens targeting lower BP goals.

FIG. 47.1, Effects of angiotensin-converting enzyme (ACE) inhibitor and calcium antagonist (CA) compared with placebo and more intensive compared with less intensive blood pressure–lowering regimens on the risks of major vascular outcomes and death. ∗ Overall mean blood pressure difference (systolic/diastolic) during follow-up in the actively treated group compared with the control group, calculated by weighing the difference observed in each contributing trial by the number of individuals in the trial. The negative values indicate lower mean follow-up blood pressure levels in the first-listed treatment groups (i.e., ACE inhibitors, CA, more). ACE-I, ACE inhibitor; CI, confidence intervals; more, more intensive blood pressure–lowering regimen; less, less intensive blood pressure lowering regimen.

More than 17,000 major CV events (a composite outcome comprising stroke, CHD, and HF events plus death from any CV cause) contributed to the overview analyses (see Fig. 47.1 ). There were significant reductions in the risk of this summary outcome measure with active treatment based on either ACE inhibitors (22%) or CCB (18%) compared with placebo, and for more intensive compared with less intensive regimens (14%). For fatal events attributable to CV or all causes, ACE inhibitor–based regimens reduced the risk of death by 20% or 12%, respectively, compared with placebo. There was also a trend toward fewer CV deaths with CCB–based regimens. However, there was no clear evidence of a reduction in risk for fatal CV events or death from any cause with regimens targeting lower BP goals.

These findings from BPLTTC have been confirmed and extended by other large-scale meta-analyses. In a report based on aggregate data from 123 studies with 613,815 randomized participants, relative risk (RR) reductions were proportional to the magnitude of the blood pressure reductions achieved, and every 10 mm Hg reduction in systolic BP significantly reduced the risk of major cardiovascular disease events (RR 0·80, 95% confidence interval [CI] 0·77 to 0·83), coronary heart disease (0·83, 0·78 to 0·88), stroke (0·73, 0·68 to 0·77), and heart failure (0·72, 0·67 to 0·78), which, in the populations studied, led to a significant 13% reduction in all-cause mortality (0·87, 0·84 to 0·91). However, no clear effect on the risk of developing renal failure was found (0·95, 0·84 to 1·07) ( Fig. 47.2 )

FIG. 47.2, Standardized effects of a 10 mm Hg reduction in systolic blood pressure. RR, Relative risk.

You're Reading a Preview

Become a Clinical Tree membership for Full access and enjoy Unlimited articles

Become membership

If you are a member. Log in here